Why Do Managers Fight Shareholder Proposals? Evidence from SEC No-Action Letter Decisions

Transcription

1 Why Do Managers Fight Shareholder Proposals? Evidence from SEC No-Action Letter Decisions John G. Matsusaka, Oguzhan Ozbas, and Irene Yi University of Southern California March 2017 New Working Paper Series No. 7 Stigler Center for the Study of the Economy and the State University of Chicago Booth School of Business 5807 S Woodlawn Ave Chicago, IL 60637

2 Why Do Managers Fight Shareholder Proposals? Evidence from SEC No- Action Letter Decisions John G. Matsusaka, Oguzhan Ozbas, and Irene Yi University of Southern California March 2017 This paper studies whether corporate managers are pursuing their own interests, or defending shareholder interests, when they fight shareholder proposals. Managers argue that shareholder proposals are harmful to firm value because they are uninformed or opportunistic, while activists argue that managerial opposition is self-interested and proposals increase firm value by counteracting managerial agency problems. We examine stock price movements following SEC noaction letter decisions to provide evidence on managerial motives. Relative to previous studies that are limited by an inability to identify precisely when investors become aware of a proposal, our new approach is to study a well-defined event date at which the SEC makes an uncertain and expressly value-neutral decision to block or allow a proposal to go forward, allowing causal estimates of the value consequences. We find that over the period , the market reacts positively when the SEC permits a proposal to be omitted from the proxy, suggesting that investors agree with managers that these proposals are value-destroying. Investors appear to be most skeptical about proposals relating to corporate governance and the elements of the E-Index, and proposals targeted at high-profit firms are viewed as particularly damaging. Comments welcome. We thank David Primo and workshop participants at George Mason University, UC- Riverside, University of Chicago, and University of Maryland for helpful comments; Amartya Bose and Elise Matsusaka for able research assistance; and USC for financial support. Contact authors at: matsusak@usc.edu, ozbas@usc.edu, irene.e.yi@gmail.com.

3 Why Do Managers Fight Shareholder Proposals? Evidence from SEC No- Action Letter Decisions 1. Introduction To what extent do managers pursue their own interests at the expense of shareholder interests? This question lies at the heart of ongoing controversies in economics and law concerning the modern corporation, and suspicion of managers is fueling a prominent reform movement that is dedicated to expanding shareholder rights. Economic theory offers two contrasting perspectives: Principal-agent theory emphasizes the self-interest of managers, and the dearth of incentives that would induce them to focus on shareholder value (Jensen and Murphy, 1990). The root problem is dispersed ownership that creates free-riding in monitoring and disciplining managers, a problem highlighted in any number of classic books and articles going back to Berle and Means (1932) if not earlier. In contrast, the competition view calls attention to market forces that pressure managers to pursue the interests of shareholders. Managers are constrained by corporate control markets (Manne, 1965), managerial labor markets, and oversight by boards of directors (Fama, 1980). The view that managers act in the interest of shareholders also has a long tradition in the law, and is the basis for the business judgment rule that creates a presumption that directors and managers are acting in the corporation s best interest. The purpose of this paper is to provide an empirical assessment of whether managers pursue their own or shareholder interests in a situation that seems ripe for conflict: when managers oppose shareholder proposals. A shareholder proposal, as the name suggests, is a proposal to change the company s practices or policies, made by one of the shareholders, that goes to a vote of shareholders at large. Shareholder proposals are an increasingly important weapon in the arsenal of corporate governance reformers. More than 16,000 proposals have been submitted at large corporations since 1997, votes have recently been held to break up the largest commercial banks and drive down executive 1

4 pay, 1 and proposals have pressured many companies to remove staggered boards, replace supermajority with majority voting standards, enhance proxy access, disclose their political contributions, and a variety of other issues. 2 Along with the surge in shareholder proposals has come a series of regulatory changes expanding voting rights and making it easier for shareholders to bring their proposals to a vote. 3 At the same time, managers continue to vigorously resist increased use of shareholder proposals, both through organizations that seek to influence regulations such as the Business Roundtable, and by seeking to exclude individual proposals from appearing in their firms proxy statement through the Securities and Exchange Commission s (SEC) no-action letter process. Managers argue that shareholder proposals are often misguided and likely to harm the firm, or are promoted by activist shareholders in order to advance their private interests. 4 Managerial efforts to shut down shareholder proposals provide an interesting testing ground because the conflict between managers and (a group of) shareholders is so direct. SEC rules allow companies to exclude a proposal from the proxy statement under certain conditions, such as if the proponent fails to demonstrate minimum stock ownership, the proposal relates to redress of a personal grievance, is vague or indefinite, 1 In 2015, shareholders voted on proposals sponsored by labor-affiliated groups to break up Bank of America, Citigroup, JPMorgan Chase, and Wells Fargo. An example on executive pay: Janus cuts CEO pay 40 percent after shareholder vote, Reuters (Kerber, 2012). 2 Examples: In 2005, only nine of the S&P 100 companies used majority voting for director elections; by January 2014, almost 90 percent of the S&P 500 had adopted majority voting (Choi et al., 2016). The number of S&P 500 companies with staggered boards declined from 300 in the year 2000 to 60 in 2013 (Harvard Shareholder Rights Project: 3 Following Dodd-Frank, the SEC modified rule 14a-8(i)(8) to allow proxy access proposals. Another example is the Delaware Supreme Court s CA, Inc. v. AFSCME Employees Pension Plan decision in 2008 that allowed shareholder proposals to alter a company s bylaws concerning decision making processes and procedures. 4 Seen from this perspective, shareholder proposals are at best a distraction to managers, and at worst harmful: proxy advisory firms may be so ill-informed that they advise shareholders to support proposals that are not in their interest (Larcker and Tayan, 2011; Larcker et al., 2015); and certain shareholders might use proposals as bargaining chips to induce managers to allocate corporate resources for the benefit of the proponent at a cost to shareholders at large (Matsusaka et al., 2016). Matsusaka and Ozbas (forthcoming) develop a model that captures these various issues. 2

5 or deals with ordinary business operations. If a company wishes to omit a proposal, it submits a letter to the SEC asking the staff to confirm that the agency will not take action against the company if it omits the proposal, called a no-action letter. Companies seek to exclude 31 percent of proposals that they receive, and the SEC grants a no-action letter permitting omission in 67 percent of its decisions. Our research strategy is to use the stock price reaction in the days surrounding the SEC s decision to infer the market s assessment of a proposal. If managers resist a proposal for responsible (value-maximizing) reasons, then we should observe a positive market reaction when the SEC allows a proposal to be omitted; if managers resist for self-interested (value-reducing) reasons, then we should observe a negative market reaction when the SEC allows a proposal to be omitted. We are not the first to study the effect of shareholder proposals on firm value, but we believe our identification strategy is a significant innovation. Previous research, beginning with Karpoff et al. (1996) almost uniformly fails to uncover evidence of positive or negative effects (Denes et al., forthcoming). 5 There is some reason to wonder, however, if existing research designs have enough power to detect valuation effects. The studies employ event-return methods, which have a good research track record, but the actual date at which a proposal becomes known to the market is difficult to determine; there is seldom a formal announcement or media coverage. In lieu of hard information about when the market learns about a proposal, most studies use either the date that the proxy statement is mailed to shareholders. However, in order to make a proposal, a shareholder must send a notice to the company at least 120 days before the proxy statement is mailed; companies must file their proxy with the SEC 10 days before mailing it; and SEC rule 14a-6(e)(1) requires the preliminary statement to be made immediately available for public inspection. So there is ample reason to expect the information to have reached the market before the proxy is mailed. Our study addresses this issue by employing a new research strategy that focuses on SEC decision dates. We argue that the existence of a proposal is known by the time of the SEC decision (by then it has been reviewed and discussed by the proponents, the company, the law firms representing both parties, and SEC attorneys), but the outcome of the 5 A notable exception, discussed below, is Cuñat et al. (2012) that uses different research methods. 3

6 decision which determines whether the proposal goes to a vote or is shut down is news to investors because it is not fully predictable. The return in excess of market movements in the days surrounding the SEC decision then sheds light on whether management s challenge is for responsible or self-interested reasons. By focusing specifically on proposals opposed by managers, as opposed to all proposals, we are able to speak to the issue of managerial incentives. Last but not least, because we have a specific event date for each proposal (as opposed to the proxy mailing approach, which assigns the same date to all proposals in a given proxy statement), we can estimate proposal-specific returns, and explore how proposal-specific features such as topic and sponsor type affect returns. We study hand-collected data on all 2,828 proposals for which a no-action letter was requested from October 2007 through the end of Our main finding is that the market responded positively to the grant of a no-action letter. The mean cumulative abnormal return ranges from 0.20 percent to 0.55 percent depending on the event window, and is statistically distinguishable from zero, meaning the market approved of shutting down these proposals. As a back-of-the-envelope calculation, if we assume that the probability of being denied a no-action letter is the sample average of 0.28, then the implied expected value from an uncertain shareholder vote on a challenged proposal ranges from percent to percent of firm value. 6 If we further assume that the probability of a proposal being approved, conditional on going to a vote, is 0.27 as reported in Cuñat et al. (2012), then the implied value consequence of a challenged proposal, if it had been approved in the end, ranges from to percent of firm value. These numbers appear to be material, and could explain the resistance of managers. In terms of the question that motivates the paper, the evidence suggests that managers are fighting these proposals for responsible rather than self-interested reasons. After establishing the robustness of the main finding, we explore the extent to which the market s response depended on a proposal s topic or the identity of its proponent. The most common topic is corporate governance, and proposals concerning compensation and social issues are common. We find a positive response to omission of corporate governance 6 In these calculations, for the sake of brevity, we ignore the expected value of withdrawn proposals. We provide a more comprehensive analysis of the implied value of proposals below. 4

7 proposals on average. The mean returns range from 0.33 percent to 0.80 percent and are statistically significant suggesting that managers were acting responsibly when they opposed these proposals. We also find some evidence that investors were skeptical of proposals that would have improved the quality of governance as defined by the G-Index and E-Index. We find no statistically significant response to omission of proposals related to compensation as a group, or social issues as a group. Recent court opinions and some scholarly evidence suggest that certain types of shareholders are more likely to bring proposals that advance their narrow interests rather than overall firm value. Labor unions and public pensions have been singled out (Romano, 1993, 2001; Schwab and Thomas, 1998; Matsusaka et al., 2016). We do not find evidence that the announcement return depended on the type of proponent, whether classified broadly as individuals versus organizations, or classified in a more fine-grained way that differentiates SRI funds, non-sri funds, labor unions, public pensions, and religious groups. When the SEC declined a no-action letter request and allowed a proposal to go to a vote, we find smaller, statistically insignificant mean returns. Given the positive mean return associated with omitted proposals, we might have expected to find a negative mean return associated with proposals that went to a vote, but the lack of statistical significance simply may be due to the 50 percent smaller sample size for proposals that went to a vote. And there is a subtler complication: the return associated with a proposal going to a vote can be positive even if the market considers the value of the proposal to be negative. This is because there is a third possible outcome once a request is made: the proposal can be withdrawn before the SEC makes a decision. This happens in 14 percent of the cases, usually at the request of the proposal s sponsor, following some sort of accommodation by management. If a side deal between an activist and management is harmful to shareholders at large, as suggested by theory and some anecdotes (Matsusaka and Ozbas, forthcoming), then the market might react positively when the SEC allows a proposal to go to a vote because it means that a harmful side deal was not struck, even if the proposal itself is harmful. To gain some insight on this possibility, the second substantive section of the paper develops a theoretical model of the no-action letter process that allows for proposals to be withdrawn. The model illustrates that while the sign of the return associated with an SEC 5

8 decision can be used to infer the market s assessment of whether the manager was pursuing shareholder interests, the sign of the return does not necessarily indicate the market s implied value of the proposal itself. Using the model, we prove a result that allows us to infer the mean implied value of challenged proposals from a regression model using abnormal returns. Using this method, we estimate a mean implied value ranging from to percent for the entire sample. Although not statistically significant, the magnitude of the estimate suggests that part of the positive abnormal return associated with SEC decisions is due to the fact that a withdrawn proposal did not occur. We then investigate for which firms proposals were most likely to be helpful and harmful. Our main finding in this regard is that proposals targeted at high-profit firms were expected to reduce firm value by 0.48 percent to 1.11 percent, depending on the event window, and controlling for topic and proponent type; while proposals at low-profit firms were expected to increase firm value. Investors may dislike proposals targeted at high-profit firms because they threaten to disrupt operations that are performing well. This squares with existing evidence that low-performing firms are most likely to be targeted by shareholder proposals (Denes et al., forthcoming). The main contribution of our paper is to shed light on the motives for managerial resistance to enhanced shareholder rights. The evidence is compatible with the view that managerial resistance is often based on a genuine concern that shareholder proposals harm firm value, particularly when it comes to high-performing firms, and is not merely a convenient rationalization in order to preserve managerial private benefits. This suggests that competitive pressures may be more effective than sometimes believed in controlling managerial agency problems. Our paper is also intended as a contribution to the literature that seeks to determine the value consequences of shareholder rights. The legal environment has been moving steadily in the direction of expanded shareholder rights, giving the impression of a consensus that enhanced rights are beneficial, yet the scholarly evidence is surprisingly inconclusive. At a general level, several studies that estimate the market reaction to expansion of shareholder rights fail to produce compelling evidence that investors value 6

9 having more rights. 7 In terms of shareholder proposals, our finding that the market approves on average when proposals are shut down suggests that many of these proposals are not helpful to shareholders. This does not imply that the process itself lacks value, only that some uses of it are harmful. A more holistic analysis would be necessary to draw conclusions about the overall value of the process, but our analysis suggests that the idea of a positive net value cannot be taken for granted; it requires evidence of concrete benefits that would offset the costs we find. 2. No-Action Letters and the Proposal Process Shareholder voting rights are rooted in state corporation law and corporate charter documents, but the proposal process itself is governed by SEC Rule 14a-8. The SEC began regulating the process in 1935 based on Section 14 of the Securities Exchange Act of 1934 that charged the agency to develop proxy regulations in the public interest and for the protection of investors. Over time, the SEC gradually developed a body of regulations that came to be collected in Rule 14a-8. 8 This rule has been amended many times over the years, most recently in Under state law, shareholders have a right to make proposals in person at a company s annual meetings. Because most shareholders do not attend the annual meeting, they cast their votes by proxy. The company is required to distribute a proxy statement prior to the annual meeting to all shareholders that in effect allows them to vote in absentia. The federal proxy access rules govern the conditions under which shareholders can require their proposals to be listed in the company s proxy statement. The proposal process begins with a shareholder proponent drafting a proposal and sending it to the company. The proposal offers a resolution to be voted on, as well as an argument in its favor. The resolution can take the form of a specific change in the 7 Negative evidence is in Akyol et al. (2012); Larcker et al. (2011); Stratmann and Verret (2012). Positive evidence is in Cohn et al. (2016) and Becker et al. (2013). Studies often report findings that appear to be sensitive to certain subsamples and event dates. 8 For histories of the development of the shareholder proposal rules, see Liebeler (1984) and Fisch (1993). For developments over the last two decades, see Bainbridge (2012). 9 In September 2011, 14a-8(i) was amended so that a company could no longer exclude proposals that would facilitate director nominations by shareholders (proxy access). 7

10 company s bylaws or it can be a request for the company to consider taking some action. The proposal must arrive at the company no later than 120 days before the proxy statement is to be mailed. The company then has the option to include the proposal in the proxy statement or the company can attempt to omit the proposal from the proxy statement by appealing to the SEC. If the company wishes to omit the proposal, it must submit a letter to the SEC no later than 80 days before the proxy statement is mailed; the letter notes that the company intends to omit the proposal and indicates the grounds for doing so. Typically, the company s letter also requests that the SEC staff respond by stating that the staff will not recommend the Commission take an enforcement action against the company if it omits the proposal, called an SEC no-action letter. If the company requests a no-action letter, the proponent is given an opportunity to respond, which may be followed by a series of responses from both parties. In most cases, if a no-action letter is issued, then the proposal is omitted from the proxy, while if the SEC declines to issue a no-action letter, the proposal appears in the proxy. Both the company and the proponent have the option of taking their case to a federal court if they disagree with the SEC s decision, which happens occasionally. Sometimes the proponent agrees to withdraw the proposal before or after an SEC decision, based on negotiations with the company. The proxy statement containing the proposal (if included) must be mailed to shareholders within a window before the annual meeting that is stipulated by state law (e.g., not more than 60 or fewer than 10 days in California and Delaware). There are many possible grounds for excluding a proposal under Rule 14a-8. Table 2 provides a summary of the procedural requirements for submitting a proposal (14a-8(b) to 14a-8(e) and 14a-8(h)) and substantive bases for exclusion (14a-8(i)). Procedural requirements include that a proponent must own stock worth at least $2,000 or 1 percent of firm value for at least one year before the meeting; may submit no more than one proposal per meeting; and the proposal and supporting statement may not exceed 500 words. The substantive bases for exclusion are wide ranging. At the most basic level, the proposal must be a proper subject for action under state law. A proposal can be excluded, among other reasons, if it would cause the company to violate a law, is false or misleading, relates to redress of a personal grievance, deals with ordinary business operations, 8

11 conflicts with a management proposal, duplicates another proposal in the proxy statement, or relates to a specific amount of dividends. It is important to note that the SEC does not judge the merits of a proposal when making a no-action letter decision. The Commission s information for companies and shareholders states: Do we [SEC] judge the merits of proposals? No. We have no interest in the merits of a particular proposal. Our concern is that shareholders receive full and accurate information about all proposals that are, or should be, submitted to them under rule 14a This means that there is no reason investors should make inferences about the merits of proposal from the SEC s decision. Table 2 reports the number of times that a given reason was the basis for a noaction letter in our sample. 11 The most common reasons for granting a no-action letter were, in order, that the proposal dealt with ordinary business operations, the proponent failed to demonstrate minimum ownership, and the company had already substantially implemented the proposal. Other common grounds for exclusion were that the proposal contained language that was false or misleading, the proposal conflicted with a company proposal offered at the same meeting, and the proposal was not submitted more than 120 days before the proxy statement is to be mailed. The ability to exclude proposals that are improper under state law is particularly important. Most state laws give the board the authority to run the company, so a proposal that mandates a particular action is often improper under state law (notable exceptions are bylaw amendments concerning decision and governance procedures). Therefore, most proposals are advisory or precatory in nature; they request or urge (or use similar phrasing asking) the company to take an action. In our sample, less than 3 percent of proposals are binding, meaning that proposals are overwhelmingly precatory in nature. 10 See Question 7 in Division of Corporation Finance, Securities and Exchange Commission, Staff Legal Bulletin No. 14 (CF), dated July 13, Companies often claim several grounds for exclusion in their letter to the SEC. If the SEC finds one reason to allow exclusion, it does not offer an opinion on the validity of the other grounds. So this count does not include all grounds for exclusion but rather those grounds that were flagged by the SEC staff. 9

12 3. Research Strategy and Related Literature The essence of our research strategy is to estimate the stock price reaction in the days surrounding the issuance of an SEC no-action letter decision. To put our strategy in perspective, it is useful to review previous work estimating the return associated with shareholder proposals. The key difference is the event date. Table 1, drawn from Denes et al. (forthcoming), lists previous studies. One approach, employed by three studies, is to estimate the abnormal return associated with publication of a media story on a proposal. Its primary limitation is lack of media coverage, leading to small sample sizes: Karpoff et al. (1996) capture 27 events; Smith (1996) studies 39 events sponsored by CalPERS; and Del Guercio and Hawkins (1999) study 102 announcements by public pension funds. The most popular approach, employed by 10 studies, is to estimate the abnormal return on the date that the company mails the proxy statement. A question with this approach is whether new information about proposals is actually revealed to the market on the proxy mailing date. As discussed above, market participants might already be informed because proposals are submitted to the company at least 120 days before the proxy statement is mailed. Also, companies are required by SEC Rule 14a-6 to file their proxy statements with the SEC 10 days before mailing, and the filed statements shall be deemed immediately available for public inspection (14a-6(e)(1)), meaning that proxy statements are publicly available 10 days before they are mailed. The usual finding of a small and statistically insignificant abnormal return on the proxy mailing date may be explained by the information having already reached the market prior to the proxy mailing date, rather than a proposal having no effect on firm value. 12 A third approach, employed by three studies, is to estimate the abnormal return on the date of the annual meeting. Because the existence of a proposal is certainly not news at the time of the annual meeting, unconditional mean returns on the meeting date are 12 Another limitation of using the proxy mailing date is that companies often have multiple proposals on the same ballot. With multiple proposals on one event date, it is not possible to isolate effects for individual proposals, and interpretation of the net effect is cloudy: if there are 4 proposals and an abnormal return of zero percent, it could mean that none of the proposals affects value, that half of them increase and half of them decrease value, and so on. Another concern is that proxy statements deliver a variety of information in addition to shareholder proposals. 10

13 negligible (Karpoff et al., 1996; Thomas and Cotter, 2007). Conditioning on the voting outcome, Karpoff et al. (1996) find a positive mean return for proposals that receive a majority of votes in favor, although the return is not estimated with precision. Cuñat et al. (2012) build on this idea by using regression discontinuity methods to distinguish proposals that narrowly pass from those that narrowly fail ; their conservative estimate is that a successful proposal produces an abnormal return of 1.3 percent on average on the meeting date. 13 The Cuñat et al. (2012) study offers the most convincing causal estimates to date but by construction it mainly considers proposals that management did not choose to oppose. In the conclusion, we discuss how our finding of a negative value for proposals opposed by managers can be reconciled with their finding. Our approach avoids some of the limitations from previous studies, although of course, it comes with its own limitations. Here we describe our identification strategy, and then we outline advantages and limitations. Our main innovation is to study the abnormal return associated with a no-action letter decision. This date is well defined and represents arrival of new information about the prospects of a proposal. In order for our estimates to reveal whether managers are acting in the interest of investors or not when they challenge a proposal, two conditions must hold: first, market participants must be aware of the existence of a proposal before the SEC makes its decision; and second, the decision must be uncertain from the perspective of investors. The first condition is plausible because prior to the decision the proposal has been seen by various officers in the target company, by the sponsors, and by multiple attorneys in the SEC, in many cases by an outside law firm employed by the company, and sometimes legal counsel for the proponents; it has also been publicly posted on the SEC s web site. Prima facie evidence for the second condition unpredictability of the decision is the fact that the SEC grants a no-action letter in only two-thirds of its decisions. An examination of the decision criteria (Table 2) also suggests the difficulty in predicting the SEC s decision. While some criteria seem black and white, such as the proposal not exceeding 500 words, 13 Because most proposals are precatory, the vote functions more as advice to managers than an approval mechanism that triggers at 50 percent. 11

14 whether a proposal violates other restrictions will not be obvious to an outsider. 14 An outsider will not know if the proponent can or cannot demonstrate minimum ownership of the stock, and the most common substantive problems the proposal deals with ordinary business operations or has been substantially implemented or is vague or indefinite are inherently subjective. While a body of precedential decisions helps to interpret these phrases and predict the outcome, there are still gray areas. In any event, if either of these two conditions does not hold, the estimated event return will be zero, which essentially biases against finding a meaningful effect. A strength of our approach is employment of a precise date at which new information arrives about whether a proposal will go to a vote. The SEC decision is communicated to the proponent and company on the decision date, and posted on the SEC web site on the day of the decision or within a few working days. Because the SEC decision is new information, the stock price reaction around the decision date offers a good opportunity to identify management s motives for resisting a shareholder s proposal. A limitation of our approach is that it encompasses only proposals that management attempted to omit from the proxy statement. This is not a problem from the perspective of testing managerial motives for resisting proposals, but it means that our estimates in the second part of the paper regarding the expected value of challenged proposals may not generalize to all proposals. Proposals opposed by managers might have different value implications than proposals that managers accept (as suggested by the evidence in Cuñat et al. (2012)). 4. Data The empirical analysis draws on three data sources. The primary data are handcollected from no-action letter files compiled by the SEC. Since October 2007, the files are published on the SEC s web site in PDF format (the information is also available in LexisNexis). Each file contains a cover letter from the SEC that identifies the company, proponent(s), and decision date; a decision letter that explains the reason for the decision; 14 Even a request to omit a proposal because it exceeds 500 words may not be as obvious as it seems. One decision in our sample concerning the 500-word limit hinged on whether CEO was one or three words. 12

15 and various letters from the company and its legal representatives and from the proponent and its legal representatives including the proposal itself. Using these files, we handcollected the decision and decision date for each case, as well as the company, proponents, and content of the proposal. Proposals were grouped into topics, and proponents were grouped into types, as discussed below. Our data run from mid-2007 through the end of Details of the data collection are reported in the appendix. Table 3 reports the number of proposals received by companies and the number that companies attempted to omit from October 2007 through the end of For the years in which the proposals and SEC data fully overlap ( ), 35 percent of proposals are sent to the SEC with a request for a no-action letter. Of the proposals that reach the SEC during the entire period, 57 percent are granted no-action letters and permitted to be omitted from the ballot; 28 percent are not granted no-action letters; and 15 percent are withdrawn or not decided. Of the proposals for which the SEC issues a decision, 67 percent are granted no-action letters. We use CRSP data to calculate event returns. We calculate the daily abnormal returns using the market-adjusted model and the Fama-French four-factor model of Carhart (1997). The length of the estimation period is 200 trading days, and we require at least 150 days with returns. The estimation period ends 10 days prior to the event date. We winsorize event window cumulative abnormal returns at 1 percent in each tail. We use multiple event windows; all of our event windows start one trading day before the noaction letter decision date and end on dates ranging from one to 10 trading days after the no-action letter decision date. Longer event windows allow for the possibility of some SEC decisions being posted with a delay. We drop an event if the window contains another event (i.e., no-action letter decision date) for the same firm in order to avoid the contamination of abnormal returns with the impact of different decisions. This process leads to a 20 percent decline in sample size because there are many cases in which the SEC makes multiple decisions for a given firm within a short window. There is seasonality in the no-action letter process; 81 percent of no-action letter decision dates are in January, February, or March. Finally, we use Compustat to obtain firm financial information. 13

16 5. Evidence on Managerial Motives A. Overall Returns Table 4 reports the mean abnormal return associated with no-action letter decisions for the full sample. The different panels report mean returns calculated in different ways, in order to assess robustness. In panel A, which contains our main estimates, abnormal returns are calculated using the Fama-French four-factor model, returns are winsorized at the 1 percent level in each tail, and decisions with another decision in the same window are omitted. Column (1) reports the mean abnormal return for decisions that granted a noaction letter, effectively killing the proposal. Returns are reported for various windows beginning one day before the decision and extending to 10 days afterwards. Glancing down column (1), the mean abnormal return ranges from 0.20 percent to 0.55 percent, and is always statistically different from zero. Investors were pleased when the SEC granted a noaction letter, consistent with the idea that managers were acting in shareholder interests when they opposed these proposals. The finding that returns grow as the window becomes longer suggests that information diffuses across the market over a week or two after the decision date. Panels B and C of Table 4 explore the robustness of this finding. In panel B abnormal returns are adjusted simply by subtracting the market return, but are otherwise calculated as in panel A. The estimates are quite similar to those in panel A: the mean return ranges from 0.24 percent to 0.84 percent, is always statistically different from zero, and grows over time. In panel C, the statistics are calculated as in panel A except that returns are winsorized at the 5 percent level in each tail. The magnitude of the mean return declines now ranging from 0.10 percent to 0.38 percent but remains statistically significant except in one case and grows with the length of the window. 15 Table 4 reveals a fairly robust pattern that investors responded positively on average to the grant of a no-action letter, that is, the market approved on average when shareholder proposals were omitted. This supports the argument that managers fight proposals for responsible reasons. 15 The findings are very similar to panel A if returns are winsorized at the 0.5% level. 14

17 Column (2) of Table 4 reports the mean return when the SEC declines to issue a noaction letter, that is, when a proposal is allowed to go to a vote. Given the finding in column (1) that the market approves when proposals are omitted, we might expect to find negative mean returns in column (2), but for the most part that is not what we find. In panel A, which we believe contains the cleanest estimates, the mean return associated with a declined no-action letter request ranges from 0.07 percent to 0.22 percent, depending on the window. None of these estimates can be distinguished from zero statistically at conventional levels of significance, and we do not observe the returns increasing as the window expands. The inability to statistically distinguish the mean from zero is due in part to the much smaller sample size for declined no-action letters compared to granted noaction letters. Examining panels B and C reveals similar findings in that the means are not consistently different from zero, and do not grow in magnitude as the window expands. A mean that is statistically insignificant admits the possibility that the true value is positive, negative, or zero, so we cannot say much here. Figure 1 presents similar evidence graphically. The figure plots the abnormal return using the Fama-French four-factor model from 10 days before to 10 days after the decision date. The red curve shows a gradual increase in the abnormal return for decisions granting a no-action letter after the decision (day 0). The gray curve shows the relatively flat return for decisions that decline to grant a no-action letter; the large confidence interval around the estimates stands out. Once a company requests a no-action letter from the SEC, there are three potential outcomes: the no-action letter is granted, the request for a no-action letter is declined, or the proposal is withdrawn by the proponent. 16 For completeness, column (3) of Table 4 reports the mean abnormal return associated with the SEC announcing that the proposal was withdrawn. Unlike the decision to grant or decline a no-action letter request, the announcement of a withdrawn proposal may not be new information to the market because the SEC is merely conveying information it received from the company, and the company only notifies the SEC after it has received a written notification from the 16 There are two other possibilities that occur very rarely: the company may withdraw its request, and the SEC staff may decline to comment. We omit these cases from our sample. 15

18 proponent. Therefore, our event date (SEC closing the file) is probably several days after the information reaches the market. Table 4 shows that the mean return associated with a withdrawal announcement is usually negative over short windows, sizably so, and statistically different from zero, but statistically insignificant over longer windows. There is some hint that the market views a withdrawn proposal as bad news. B. By Topic of Proposal The topic of proposals varies widely, ranging from corporate governance issues to executive compensation to social issues. Different types of proposals may have different effects on firm value. To get a sense of how investors perceive different types of proposals, this section reports the mean returns associated with different proposal topics. Figure 2 provides some context by showing the number of proposals grouped into three broad categories: compensation, corporate governance, social issues. The appendix describes the category definitions in detail. As can be seen, corporate governance proposals are the most popular subject. Figure 2 also shows the fraction of each group that is withdrawn, granted a 16

19 no-action letter, or had a no-action letter request declined. Withdrawal and SEC decision rates vary somewhat across proposals. Panel A of Table 5 breaks out the mean returns separately for three broad groups of issues: compensation, corporate governance, and social issues. Throughout Table 5, the estimates are based on abnormal returns calculated using the Fama-French four-factor model and winsorized at the 1 percent level in each tail, and windows with more than one event are omitted. Management s decision to oppose proposals concerning their own compensation deserve particular scrutiny due to the inherent conflict of interest. For compensation proposals, Table 5 shows that the mean return associated with the grant of a no-action letter is positive, ranging from 0.05 percent to 0.57 percent, but not statistically significant. Similarly, for social issues, the mean return associated with a no-action letter is usually positive, but never statistically different from zero. The interesting finding is for corporate governance issues, for which the mean return is always positive and statistically different from zero at the 5 percent level. The means range from 0.33 percent to 0.80 percent, and grow with the window size. 17

20 Panel B of Table 5 reports the mean returns for a different grouping of issues. Panel B1 focuses on proposals that would improve governance according to the G-index, and Panel B2 focuses on proposals that would improve governance according to the E-index. Both indexes are intended as summary measures of the quality of a company s governance provisions; they are correlated with a variety of performance metrics and enjoy some popularity among reformers (Bebchuk et al., 2009). The G-index has 24 elements, listed in the appendix., and the E-index has six elements: board declassification, majority voting on bylaw amendments, majority voting for mergers, majority voting for bylaw amendments, limits on golden parachutes, and removal of poison pills. The estimates in Panel B include all proposals that correspond to one of the elements of an index. Somewhat surprisingly, Panel B of Table 5 shows a consistently positive mean return when the SEC shut down proposals that would have improved governance according to the indexes. The mean return associated with omitting G-index proposals ranges from 0.45 percent to 0.60 percent, and is statistically significant for three of four reported windows. The mean return associated with omitting E-index proposals ranges from 0.68 percent to 1.03 percent, and is statistically different from zero in three of four windows. The mean return associated with the SEC declining to issue a no-action letter is smaller, usually negative, and never distinguishable from zero at conventional levels of statistical significance. There is enough noise in the estimates to be cautious about drawing conclusions, but the estimates seem to be suggesting that investors believed that increases in governance quality as measured by the G-Index and E-Index reduced firm value. The estimates in Table 5 suggest that investors were uncomfortable with at least some corporate governance proposals. It is natural to wonder precisely which proposals were the main cause of concern. Table 6 reports the mean returns associated with SEC decisions for the six most common specific types of governance proposals in our sample: special meetings, majority voting, independent chair, proxy access, board declassification, and written consent. Data mining is a risk in this sort of exercise because the number of proposals of any specific type can be small and some means are likely to pass critical values for statistical significance simply by chance, so we are hesitant to draw strong conclusions. The basic patterns are the following: 18

21 Special meetings. The most common type of proposal expands the right of shareholders who meet certain minimum ownership conditions to call a special meeting of shareholders. The mean return is positive from omitting these proposals and negative from allowing them, but not statistically significant. Majority voting. The second most common type of proposal requires a majority vote for corporate elections (for directors, bylaws, or control transactions). The move from plurality to majority voting for corporate directors has been one of the most successful planks in the reform platform over the last decade (Choi et al., 2016). The mean return associated with either type of SEC decision on such proposals is positive (with one exception) but not consistently different from zero statistically. Independent chair. The next most popular type of proposal prohibits the same person from serving as CEO and chair of the board. The mean returns are inconsistently signed and never statistically significant. Proxy access. Perhaps the strongest evidence concerns proposals to enhance proxy access. Such proposals allow shareholders who meet certain ownership conditions to nominate candidates for the board. Reformers won an important victory in 2010 when the SEC adopted new rules allowing shareholder proposals on proxy access. 17 The mean return associated with omitting such proposals is consistently positive, ranging from 0.24 percent to 1.91 percent, and statistically different from zero for the longer windows. The evidence is not conclusive, but suggests that investors view proxy access proposals with some skepticism. This comports with evidence based on legislative histories finding negative or zero market reactions to enhanced proxy access (Akyol et al., 2012; Larcker et al. 2011; Stratmann and Verret, 2012). 17 At the same time, the SEC also adopted a rule mandating the shareholder right to nominate directors, but that rule was vacated by the D.C. Court of Appeals in Business Roundtable v. SEC,

22 Board declassification. Proposals to declassify the board would require all directors to be elected on an annual basis. The alternative staggered elections make it more difficult to change control of the board. Several studies have documented a negative correlation between staggered boards and firm value, but recent causal estimates produced mixed findings, or suggest that staggered boards increase value (Cremers et al, forthcoming). The mean return associated with omission of declassification proposals is large and positive in all windows, ranging from 1.17 percent to 2.04 percent, statistically significant at the 10 percent level in three of four windows. The mean return from allowing declassification proposals is positive, but not statistically significant. Written consent. Finally, the mean return associated with proposals allowing shareholders to act by written consent is positive, and never statistically different from zero. To summarize, the evidence suggests that the market s concern with shareholder proposals is primarily centered on corporate governance proposals. However, we do not have enough data and precision to drill down very deeply into how investors view specific types of proposals. 18 C. By Type of Proponent Just as investors may view some proposal topics more favorably than others, their assessments may also be colored by the identity of the proponent. Because the proponent might be asymmetrically informed about the consequences of a proposal, theory suggests 18 We also calculated but do not report mean returns associated with proposals that ask companies to reveal their political contributions. Such proposals have been popular recently. Min and You (2015) show that such proposals are targeted at companies with a history of donating to Republican candidates, suggesting they may have a political motivation rather than a value-enhancing goal. The mean returns suggest that investors are skeptical of such proposals, but the magnitudes are not large and the means are never statistically different from zero. 20

23 that investors will base their beliefs about its value implications in part on the identity of the sponsor. Here we examine the mean return associated with SEC decisions, by type of proponent. 19 It is not obvious ex ante which proponents are suspect and which are not, so our estimates should be seen as exploratory. Labor unions and public pensions have been singled out by researchers and in court decisions for potentially using the proposal process to advance private goals that do not maximize value, such as benefits for union workers, which might make investors skeptical about their proposals. But on the other hand, unions may have useful information about a company s operations, putting them in a good position to recognize value-enhancing opportunities. Another group that has been the subject of much discussion is hedge funds. Hedge funds have been lauded for their focus on shareholder value, and substantial evidence exists that activist campaigns increase value (Brav et al., 2015). Others have expressed concern that hedge funds might be short-term oriented, pressing the firm to generate short-term cash distributions at the expense of long-run value (Anabtawi, 2006). Figure 3 reports the number of proposals in our sample by type of proponent. Individuals make by far more proposals than any other sponsor category, with 1,477 in our sample. This is not simply because proposals from individuals are more likely to be challenged: for the entire ISS Proposals database over , 38 percent of proposals are sponsored by individuals. Some of these proposals originate with persons who make a one-time proposal on a specific issue, but most come from a small number of so-called corporate gadflies who make dozens of proposals every year. The most active gadflies in our sample are Gerald R. Armstrong, John Chevedden, husband-wife team James McRitchie and Myra Young, and father-son team William Steiner and Kenneth Steiner. Also mirroring the pattern in the ISS Proposals database, we find very few proposals (a total of 45) from hedge funds (included under Fund (non-sri) ). This is because shareholder proposals are not an important element of hedge fund activism in general; hedge funds are more focused 19 Another (indirect) way to assess how other shareholders view a proposal is through their votes: the evidence connecting votes to the identity of the sponsor is mixed and inconclusive (Thomas and Martin, 1998). 21